sydepsystem design atlas

IoT Telemetry Platform

Ingest device messages, route them to storage and alerting in near real time.

AdvancedIoTPlatform
DevicesEdge & IoTIoT GatewayEdge & IoTTelemetry StreamMessagingRaw TelemetryStorageThreshold AlertsObservability

Devices send telemetry to an IoT gateway, which fans the messages into a stream, stores raw payloads, and alerts on thresholds or out-of-band conditions so operators can react quickly.

When to use it

  • High-volume device telemetry arrives continuously from many places and must be processed near-real-time
  • Both raw telemetry retention and threshold-driven alerts matter

Trade-offs

  • The gateway is often the bottleneck under bursty device writes and needs careful capacity planning
  • Device identity and secure provisioning become as important as the data pipeline itself

Components used

Device GatewayIoT HubEvent StreamObject StorageAlerting & On-call

How it works

  • Devices connect over a lightweight protocol suited to unreliable, low-bandwidth networks and publish readings to a broker.
  • An ingestion tier authenticates devices, validates payloads and writes to a time-series store, while a parallel rules path evaluates alert conditions in near real time.
  • Device identity and certificate lifecycle are first-class concerns, since devices vastly outnumber human users.

Used in the wild

  • Industrial sensor monitoring and predictive maintenance.
  • Connected consumer devices reporting status and usage.
  • Fleet and asset tracking.

Good to know

  • Devices have clocks that drift or reset, so payload timestamps cannot be trusted. Recording both device time and ingest time is what makes the data debuggable later.
  • The reconnect storm is the classic outage: a network blip drops a million devices, all of which retry immediately and simultaneously. Randomised backoff in device firmware is the only real defence, and firmware is the hardest thing to change.